How2QA 视频+语言数据集

How2QA 是一个视频+语言学习框架数据集。数据集将同一组选定的视频片段呈现给另一组 AMT 工作人员进行多选择问答标注。每位工作人员被分配一个视频片段，并被要求根据四个准备好的回答（一个正确答案和三个分散注意力答案）写一个问题。对工作人员隐藏视频叙述，以确保收集的问答对不受字幕的影响。数据集包含从 9035 个视频中选择的 22000 个 60 秒剪辑片段和 44007 个问答对。

How2QA is a video-and-language learning framework dataset. The dataset presents the same set of selected video clips to another group of AMT workers for multiple-choice question answering annotation. Each worker is assigned one video clip and asked to generate a question based on four prepared responses (one correct answer and three distractors). The video narration is concealed from the workers to ensure that the collected QA pairs are not affected by subtitles. The dataset contains 22,000 60-second clips selected from 9,035 videos and 44,007 question-answer pairs.

搜集汇总

背景与挑战

背景概述

How2QA 是一个用于视频与语言学习的多模态数据集，它基于从9035个视频中选取的22000个60秒剪辑片段构建。该数据集通过让工作人员在隐藏视频字幕的条件下，为每个片段编写问题并准备答案，共包含了44007个问答对，旨在支持视觉问答和视频理解等任务。